[paper] Generative AI for Analog IC Design

D. Noori Zadeh and M. B. Elamien

Generative AI for Analog Integrated Circuit Design: Methodologies and Applications

in IEEE Access, vol. 13, pp. 58043-58059, 2025

DOI:10.1109/ACCESS.2025.3553743. 

* Department of Electrical and Computer Engineering, McMaster University, Hamilton, ON, Canada

Abstract: Electronic Design Automation (EDA) in analog Integrated Circuits (ICs) has been the focus of extensive research; however, unlike its digital counterpart, it has not achieved widespread adoption. In this systematic review, we discuss recent contributions in the last five years, highlighting methods that address data scarcity, topology exploration, process-voltage-temperature (PVT) variations, and layout parasitics. Our goal is to support researchers new to this domain by creating a comprehensive collection of references and practical application guidelines. We provide a methodological review of state-of-the-art machine learning (ML) approaches, including graph neural networks (GNNs), large language models (LLMs), and variational autoencoders (VAEs), which have been successfully applied to analog circuit sizing tasks. To the best of authors' knowledge, this is the first review to comprehensively explore the application of generative AI models in analog IC circuit design. We conclude that future research could focus on few-shot learning with domain-adaptation training of generative AI methods to simplify the design tasks such as human-tool interaction or guided design space exploration.

FIG.1: Analog design automation flow, focusing on circuit-level automation.

Acknowledgements: This work was supported by the Natural Sciences and Engineering Research Council (NSERC) of Canada through its Discovery Grant (DG) Program under Grant RGPIN-2024-06826.